AIRBNB x RENTA SEGURA

FINAL PROJECT

Table of Content

1) Introduction

2) Our Dataset

2.1) Curation of the dataset (2)
2.2) General Overview (1)
2.3) Statistical Analysis (2)

3) Possible Solutions

3.1) Per Neighborhood (3)
3.2) Per Market Player (1)
3.3) Per Usage Ratio (2+)

4) Conclusion

Introduction

For our project, we went to analyze a dataset containing information about Airbnb listings. As the proposal submitted already shows, we found a sharp decline in the short-term rental business due to a dizzying reduction in tourism, which prompted us to look for possible recovery solutions. Our main idea is to implement the project already started by the municipality of Lisbon "Renda Secura" to the apartments in Aibnb. In this way, in fact, we could resume a currently stagnant economy by giving a fixed income to the owners of the apartments and repopulating the city with its citizens.

We will alternate parts of storytelling with parts of dataset analysis and its consequent outputs during the whole Notebook course. The work is structured in different phases. The first part of the work includes the curation of the dataset, a general overview of rentals in Lisbon, and a statistical survey of variables.

We wondered, then, what were the possible solutions to the market stagnation. We then identified three possible ways to focus on. First, we analyzed the listings based on the different neighborhoods, looking for those most affected. Then we went to see which were the major players in the market so that we had the opportunity to enter into multiple contracts and to obtain subsidized prices. Finally, we tried to identify a small number of listings through the usage ratio, prioritizing those vacant apartments in the past year (dates to be entered).

OUR DATASET

2.1 Curation of the dataset

Having a very large dataset available, as first thing we picked only the data needed for the analysis.Then, we made adjustments to the columns that required it (e.g. we rounded up the number of bathrooms and corrected typos). We noted the presence of price outliers that muddled the distribution. Consequently, as shown in graph1, we decided to consider only prices up to 500€ per night, excluding 1% of listings.

As our project is the implementation of Renta Segura to the houses in Lisbon used for short term rentals, we created a copy dataset containing only the listings of the city without affecting the initial dataset.

2.2 GENERAL OVERVIEW

This is Joao:

Joao, his wife Jannina and his son José are expecting a fourth family member: Joaquim. Due to the fact that the current apartment will soon become too small, Joao is looking for a new home. The family wants to stay in Joaos and Joanas hometown: Lisbon . They ask their current landlord Jair for help, because he owns several apartments in the city. However, Jair immediately rejects Joaos and Joanas request: “Ahahaha, no way. I am renting my properties to tourists, they have more money than you.”

In the recent years, the city of Lisbon welcomed more and more tourists each year. Joao likes to meet people from other countries, but with their high wages from rich countries the tourists occupy more and more housing spaces in the city. It turned out that they do not only stay in hostels but also rent whole apartments. Most of these short-term rental agreements are closed via a platform called Airbnb. Let’s see what kind of apartments we can find on Airbnb in Lisbon and where they are:

Joao, desperate, decides to turn to his friend Frederico, who is following the development of the Renda Segura project.

-Joao: "With the birth of Joaquim, we will definitely need more space. I tried to ask Giacomino, my current landlord, but I can't afford anything I'm looking for with the prices going around. The only hope I have left is the project that you have started in common. With the remaining two-thirds of my salary, I will, in fact, be able to ensure a dingy life for my wife and both of my children."

-Frederico: "I fully understand your problem, my friend. After years working in real estate, I can assure you that the number of rooms, the size and the area in which the house is located are the variables that most influence prices. Since we have occupied all the available solutions, we are looking for new houses to rent to Lisbon citizens like you. In this regard, we are developing research to identify possible dwellings that can be converted from short term to medium-long term rentals. To make you understand better, I'll show you some graphs that our analysts have created through the dataset of Airbnb listings in Lisbon. In the first one, confirming the regression I just explained, there is a continuous growth of prices based on the rooms. Instead, in this second one, we divided the different neighborhoods according to the price per night, showing the average across the city. In the latter, we have calculated the density of listings by area to see which ones have more availability."

-Joao:" So have you identified any neighborhoods to start with yet?"

-Frederico: "Our team was able to create a map of Lisbon divided by neighborhoods that shows for each one some information such as: average price, area, and number of listings available. In this way, we got an idea of what could be the most plausible solutions. However, we need to figure out which apartments have been vacant for the longest time and which owners would be interested in our project."

2.3 STATISTICAL ANALYSIS

In this section, we analyzed some variables in our dataset to verify the feasibility of the project and how Airbnb's business is structured. For this reason, we decided to suspend our storytelling, as we do not think poor Joao would be interested in a purely quantitative analysis.

The graph below shows other data about the listings price. In particular, it shows the density of the flats related with the price. it emerges that there is a particular aggregate of houses with price between forty and eighty. In the same range we recorded a spike of 0.016.

As already mentioned in the curation part, we've focused only on a part of the entire dataset. The graph below, indeed, just to have an idea of the distribution for each column we used, shows the ones we took in consideration. Furthermore we analyzed the correlation between them rsppresenting the ones we retain relevants ( corr > 0.3 & corr < -0.3)